It’s Time to Rethink Data Centers for the IoT and Big Data

There will be 6.4 billion connected things in use worldwide in 2016, according to market research firm Gartner forecasts, with 5.5 million new things getting connected every day. Doing a bit of math on that prediction, by the end of next year there could well be over 8 billion Internet of Thing devices generating data from people, places, and things.

Imagine all the data generated by those devices – surely there will be zettabytes or even yottabytes of information that must be stored and analyzed. The big question that is looming on everyone’s mind is this: Will data centers be able to handle all that data?

No, they won’t, says Manny Linhares, director of strategy for the data communications division of Legrand, a provider of products and systems for electrical installations and information networks. Linhares, a former mechanical engineer, is now providing IoT expertise to companies that are building out networks and data center infrastructures. And he has some interesting insights on how the IoT is forcing companies to rethink data centers and the management of big data.

“The world is just not going to be able to hold all the data,” Linhares says. He uses the increasing intelligence of lighting in buildings as an example. “Soon, Cisco will be one of the largest players in LED lighting. Who would have thought that a network giant would become a foremost leader in intelligent LED lighting?”

With Cisco networks powering up lighting, data lines will now become low-voltage power sources. “Rather than mechanically wired, lights in a building will be connected to the network via IoT devices. Software will be turning them on and off, rather than an electrical wire and traditional light switches typically used to control high-voltage power,” Linhares says. IoT lighting switches will be able to power lights on and off, change their color, adjust to moods, dim, and sense natural lighting. All of this will be over a network connection without the need for a separate power supply.

The good news is that there will be tremendous energy savings. But here’s the catch. As buildings become more intelligent with IoT devices, Linhares says data centers will need to respond to new multiple connections and massive amounts of data – maybe 10 to 20 times what they are managing today. Data is being generated at rates that never have been seen before, but where is all the data going to go?

“As a director of strategy, I get asked quite often the question, ‘are data centers going to get bigger because of IoT?’ My answer is almost instantaneous,” Linhares says. “And it’s ‘I don’t know.’ What I do know is that data centers will not be able to handle everything.”

To ease the pressure on data centers and data storage while preserving the benefit of IoT intelligence, Linhares has two suggestions. First, he recommends that companies look to decentralize their data center, using the concept of fog computing and gateway devices to capture IoT data locally rather than sending it all directly to the data center.

“With remote gateway devices near IoT devices, data centers can be flattened so they don’t need to get bigger,” Linhares says. “Think about a hospital. There could be a gateway device on a surgery wing that collects data from hundreds of IoT sensors and devices in the operating rooms. From the gateway, one cable line can run to the data center. This kind of one to ‘n’ connection can create enormous efficiencies both in networking and energy.” For instance, rather than adding a $100,000 switch in the data center, an organization could add several $50-500 gateways in several locations throughout the premise.

The next step Linhares recommends is for companies to scrutinize the data it captures and determine what is really of value. “It’s time for companies to rethink how data is being captured. For instance, at an ATM, a security camera doesn’t need to capture hours of inactivity – it only needs to provide intelligence on a change in activity or motion,” Linhares says. “Petabytes of data is generated when no one is there – and that data has no value. But if the sensors and built-in intelligence are cued up to show a change from inactivity to activity, then IoT images can capture data that is of real value.” With this approach, companies can not only increase the value of IoT data, but also save tremendously on storage.

Both recommendations – the decentralization of the data center and greater scrutiny over which IoT data to capture – are different ways to look at how things can be done today. But soon, with 8 billion connected devices in the world, companies won’t have much choice. They will have to do things differently.